Add location-anchored comments to images as agent context

[Felippen]

I'd like to click an image in T3 Code and add comments to specific points or areas, similar to the ChatGPT desktop app shown in the screenshots.

Suggested workflow

1. Open an attached or displayed image and choose Comment from a toolbar.
2. Click a point or select an area, then write a comment. Show numbered markers for each comment.
3. Collect the comments alongside the image in the composer, with options to edit or remove them.
4. Send the image and comments to the agent, preserving where each comment was placed so the agent understands exactly what it refers to.

For example, I could point to a button in a screenshot and say "Move this below the heading" without having to describe its location.

This request is for location-based feedback on images, rather than a general image editor, cropping tools, or a sketching pad. It should work with image attachments, independently of browser-preview annotations.

Reference screenshots

The image toolbar and Comment action

A numbered comment placed on the image

The comment collected beside the image in the composer

[juliusmarminge]

Note

Grok responding on behalf of Julius.

This looks like the same ask as #19126 (annotate pasted images with numbered location comments before sending). Closing as a duplicate of that thread — please add any ChatGPT-style UX details there.

Related: #14812 (cite/crop a region when replying to agent images).